On Tue, Sep 26, 2017 at 08:17:59PM +0000, Bart Van Assche wrote: > On Tue, 2017-09-26 at 22:54 +0800, Ming Lei wrote: > > On Tue, Sep 26, 2017 at 02:29:06PM +0000, Bart Van Assche wrote: > > > On Tue, 2017-09-26 at 17:15 +0800, Ming Lei wrote: > > > > No, if you only address issue on MD device, that is definitely not > > > > alternative of my patchset. > > > > > > A clarification: my patch series not only fixes suspend and resume for md-on-SCSI > > > but also for all other scenario's in which resume locks up due to no tag being > > > available when the SCSI core tries to execute a power management command.> > > I do care about if this patchset can fix non-MD cases, like > > btrfs/raid, or transport_spi, both are real reports. Could you > > make sure if your patchset can cover this non-MD devices? > > Yes, I'm sure this patch series covers non-MD devices. > > It is easy to see that this patch series also covers SCSI parallel domain > validation. Now I don't believe this patchset can do that, see my comment on your patch 5: https://www.mail-archive.com/linux-block@xxxxxxxxxxxxxxx/msg13624.html More worse, your patch introduces a new I/O hang inside SCSI quiesce. > > Regarding BTRFS RAID: you should know that filesystems are responsible for > implementing freeze / thaw support themselves. I think a filesystem must > surround metadata operations by sb_start_write() / sb_end_write() and that > it optionally can provide freeze_super / thaw_super callbacks in > struct super_operations. In the BTRFS changelog there are several fixes > with regard to freeze / thaw so I think this aspect of BTRFS has been tested. > > Bart. -- Ming
